Yes, just be aware that MATLAB’s imresize has anti-aliasing enabled by default:
imresize(A,scale,'bilinear')
vs. what you would get with cv::resize(), which does not have anti-aliasing:
imresize(A,scale,'bilinear','AntiAliasing',false)
And as Amro mentioned, the default in MATLAB is bicubic, so be sure to specify.
Bilinear
No code modifications are necessary to get matching results with bilinear interpolation.
Example OpenCV snippet:
cv::Mat src(4, 4, CV_32F);
for (int i = 0; i < 16; ++i)
src.at<float>(i) = i;
std::cout << src << std::endl;
cv::Mat dst;
cv::resize(src, dst, Size(0, 0), 0.5, 0.5, INTER_LINEAR);
std::cout << dst << std::endl;
Output (OpenCV)
[0, 1, 2, 3;
4, 5, 6, 7;
8, 9, 10, 11;
12, 13, 14, 15]
[2.5, 4.5;
10.5, 12.5]
MATLAB
>> M = reshape(0:15,4,4).';
>> imresize(M,0.5,'bilinear','AntiAliasing',true)
ans =
3.125 4.875
10.125 11.875
>> imresize(M,0.5,'bilinear','AntiAliasing',false)
ans =
2.5 4.5
10.5 12.5
Note that the results are the same with anti-aliasing turned off.
Bicubic Difference
However, between 'bicubic' and INTER_CUBIC, the results are different on account of the weighting scheme! See here for details on the mathematical difference. The issue is in the interpolateCubic() function that computes the cubic interpolant’s coefficients, where a constant of a = -0.75 is used rather than a = -0.5 like in MATLAB. However, if you edit imgwarp.cpp and change the code :
static inline void interpolateCubic( float x, float* coeffs )
{
const float A = -0.75f;
...
to:
static inline void interpolateCubic( float x, float* coeffs )
{
const float A = -0.50f;
...
and rebuild OpenCV (tip: disable CUDA and the gpu module for short compile time), then you get the same results:
MATLAB
>> imresize(M,0.5,'bicubic','AntiAliasing',false)
ans =
2.1875 4.3125
10.6875 12.8125
OpenCV
[0, 1, 2, 3;
4, 5, 6, 7;
8, 9, 10, 11;
12, 13, 14, 15]
[2.1875, 4.3125;
10.6875, 12.8125]
More about cubic HERE.